1. Field
This invention is in the field of fluid flow control and, more specifically the field of high accuracy flow control such as required for, e.g., semiconductor processing, plat panel display fabrication, solar cell fabrication, etc.
2. Related Art
Metering the mass-flow rate of a gas is important to many industrial processes. In the case of the semiconductor industry, metering must be especially accurate, because deviations in the flow rate of only several percent can lead to process failures.
Mass flow is the result of a pressure gradient existing in a system. As long as no external work is done on the system, mass will flow from areas of high pressure to low pressure. This is the working principle in all flow control devices. In order to control the rate of flow from the high pressure region to the low pressure region, a flow restriction is used. The flow restriction is positioned such that all flow in the system must pass through the restriction. Depending on the characteristics of the restriction, mass flow rate through the flow control device is a function of some or all of the following: dimensions of the flow restriction, the magnitude of the pressures both upstream and downstream of the flow restriction, the temperature of the system, and the physical properties of the gas, such as density and dynamic viscosity. The flow rate can be controlled by varying one or more of these parameters. In general, the physical properties of the gas and the temperature of the system are difficult to change or control, so flow is controlled by varying the pressures in the system, the dimensions of the flow restriction, or both.
The industry-standard flow control device is a mass flow controller (MFC) containing a flow restriction in the form of a valve that can be partially opened to allow increased flow or partially closed to decrease flow. The opening of the valve is controlled by a closed loop feedback circuit that minimizes the difference between an externally provided set point and the reading from an internal flow measuring device. The flow measuring device uses a thermal sensor with two resistance-thermometer elements wound around the outside of a tube through which the gas flows. The elements are heated by applying an electric current. As the gas flows through the tube, it picks up heat from the first element and transfers it to the second element. The resulting temperature differential between the two elements is a measure of the mass flow rate of the gas. In the newer, pressure insensitive MFCs, a pressure transducer is included between the thermal sensor and the control valve to account for the effects of changing pressure on flow.
A consequence of the thermal sensor flow measurement used in the MFC is that accurate flow control requires regular calibration of the device. Without regular calibration, the actual flow rate through the MFC can drift to unacceptable values due to errors in the flow measuring device. This calibration is often performed by flowing gas through the MFC into or out of a known volume and measuring the pressure rise or drop in the volume. The actual flow rate can be determined by calculating the rate of pressure rise or drop and using established pressure-temperature-volume gas relations. This type of measurement is known as a rate-of-rise calibration.
The rate-of-rise flow calibration is based on primary flow measurements, and is therefore a primary calibration standard—that is, flow is determined only by measurements of mass, pressure, volume, and time. There are only three types of known primary flow measurements: gravimetric, measuring the change in mass over time; volumetric, measuring the change in volume at constant pressure over time; and rate-of-rise, measuring the change in pressure at constant volume over time. All other types of flow measurement are secondary measurements and must be calibrated to a primary measurement.
Another method of metering the flow rate of a gas is to vary the pressure of the gas upstream of a critical orifice. The volume-flow rate of a gas through a critical orifice at constant temperature is independent of the upstream or downstream pressure, provided that certain pressure requirements are met, e.g., the upstream pressure is twice that of the downstream pressure. By controlling the density of the upstream gas, which is proportional to pressure, the mass-flow rate through the critical orifice can be controlled.
In this type of flow control, the pressure is controlled using a control valve in a closed loop control circuit with a pressure transducer positioned between the control valve and the critical orifice. The control valve is opened or closed to maintain a specified pressure upstream of the critical orifice. Mass flow rate is determined from the pressure upstream of a critical orifice and the established characteristics of the critical orifice. Accurate flow metering, therefore, is dependent not only on the performance of the pressure controlling system, but also on the mechanical integrity and stability of the dimensions of the orifice. Since the orifice is susceptible to being restricted with particulate contamination or eroded with reaction by the gases flowing through it, it is desirable to calibrate the pressure-flow relationship on a regular basis. This is performed using the same rate-of-rise measurement that is used for the MFC.
Both of the above mentioned methods control mass flow using a closed loop control scheme in which mass flow is ultimately the result of a pressure gradient acting across a flow restriction. Viewed as a control system, the output variable of these devices is mass flow, and the input variables are pressure and flow restriction characteristics.
In the case of the MFC, it controls the dimensions of the flow restriction based on a second-order measurement of mass flow rate. The actual dimensional characteristics of the flow restriction are unknown, but can be adjusted proportionally to increase or decrease flow restriction as desired. In terms of the process variables, flow restriction and pressure, only pressure is observable by the device (for the pressure-insensitive MFCs), and only the flow restriction can be controlled.
The critical orifice device controls flow by monitoring and controlling the upstream pressure while maintaining presumably constant flow restriction characteristics. The critical orifice device does not monitor or control the characteristics of the flow restriction beyond assuming they are constant. In terms of process variables, pressure is both observable and controllable by the device, while flow restriction is not controllable or observable. It is true that without any external influence, the characteristics of the flow restriction should not vary with time; however, in operation, the possibility exists for either chemical or mechanical perturbation of the flow restriction. This type of perturbation cannot be measured by the system, and therefore, cannot be corrected without the aid of an external calibration.
The shortcomings of both of these flow control schemes, especially the need for external measurements for calibration and detection of faults, illustrate why an improved flow control scheme is desirable.
A key requirement of a flow control device that is able to detect faults in its operation as well as to correct those faults through self-calibration is that there be a sufficient number of process variables that are observable and controllable. For both types of flow control devices described above, which together comprise the vast majority of flow control devices used in the semiconductor industry, there are not sufficient process variables to accomplish these tasks.
In the present invention, these process variables are added by implementing a control valve that is designed to provide a highly accurate and repeatable mapping between its position and its flow restriction characteristics, and is able to achieve a very accurate measurement and control of its position.
If the flow restriction can be controlled and also measured, the only additional input necessary to control the flow rate is knowledge of the pressure gradient acting on the flow restriction, because flow conductance is a knowable, repeatable function of the flow restriction dimensions. This control scheme is similar to that of the critical orifice device, except the static flow restriction is replaced with one that is variable and measureable.
The benefits obtained through use of this controllable valve will depend on the type of flow control device in which it is implemented. For the critical orifice flow control device, substitution of the controllable valve for the critical orifice will remove the uncertainty of any change to the dimensions of the critical orifice. For the thermal sensor MFC, since the combination of the pressure transducer and controllable valve provides a known flow rate, this flow rate can be checked against the flow rate measured by the thermal sensor, where any discrepancy is noted as a fault.
Neither of these flow control device types, however, allows self-calibration during operation. For that capability, the flow control device must incorporate a primary flow measurement as an integral part of its operation. Incorporating this type of control valve into the flow monitoring system shown in FIG. 1 yields a flow control device that is highly accurate and self-calibrating. Initially, the position of the control valve 108, which determines the flow restriction, would be controlled based on a rate-of-drop flow measurement carried out with valve 106 fully closed. Following the rate-of-drop measurement, valve 106 would be opened and flow would be controlled by adjusting the position of the control valve based on the pressure in the system as measured with pressure transducer 112.
A flow restriction with measurable, controllable dimensions is a key piece to a greatly improved flow control scheme, which can ultimately lead to a flow control device that is self calibrating and does not rely on secondary flow measurements. The technological challenges in making this type of control valve that is accurate enough for the semiconductor industry become apparent from an order-of-magnitude estimate of the precision required. The mass flow accuracy currently required for semiconductor processing equipment is +/−1%. In general, flow must be controlled between 1 and 10,000 sccm (standard cubic centimeters per second), and the pressure difference between the flow restriction inlet and outlet is typically between 20 and 150 psi (pounds per square inch). If we imagine an illustrative flow restriction in the form of a rectangle with a static width of 10 mm and a static length of 1 mm, with a 60 psi pressure drop across the restriction, the height must be adjusted to 0.8 um to allow flow of 1 sccm. Performing error propagation with the specification of +/−1%, the height must be controlled to within +/−1.1 nm.
In fact, except for high precision metering valves, which are not well suited for the cost, space, cleanliness, and reliability requirements of semiconductor manufacturing, very little work has been done to implement control valves with measureable and controllable restrictions. U.S. Pat. No. 6,761,063-B2, entitled “True Position Sensor for Diaphragm Valves,” uses “a thin conductive member disposed between the diaphragm membrane and the actuator” to measure the capacitance between this conductive member and the valve body. The capacitance value provides an indication of the distance between the conductive member and the valve body, which gives an indication of the distance between the diaphragm and the valve body, which then gives an indication of the amount of valve opening. This type of assembly, with all of these separate parts, will at best provide control to a precision of approximately 0.001 inch, which is approximately 25,000 nm. In addition, the inventors note that the capacitance will change when fluid flows through the valve, thus making this approach even less appropriate for measuring the characteristics of the flow restriction to the level of accuracy needed in the present invention.